Morning Briefing

Multi-source AI orchestration, daily production use

Problem

Every morning I was doing the same routine: check the weather, open my calendar, scan three news sites for AI updates relevant to my brands, and try to remember what I shipped yesterday. It took 10–20 minutes before I started working.

The worst part was not the time. It was that I had no system connecting what happened yesterday to what I should do today.

Stack

Two connected Make.com scenarios power the system. A feeder bot called Claude Sessions Sync runs every hour, pulling session logs from Notion, iterating through entries, and logging them into Google Sheets. The Morning Briefing scenario runs once daily at 08:00, pulling from that same sheet plus a weather API, Google Calendar, and three RSS feeds. Claude API synthesizes everything into a single structured Telegram message.

Make.comClaude APINotion APIGoogle SheetsGoogle CalendarWeather APIRSSTelegram Bot API

Flow diagram

Layer 1 — Claude Sessions Sync (every hour)

Notion → Iterator → Google Sheets → Notion

Make.com scenario for Claude Sessions Sync — hourly Notion to Google Sheets pipeline

Layer 2 — Morning Briefing (daily 08:00)

Google Sheets → Text Aggregator → Weather API → Google Calendar → RSS ×3 → Set Variable → Claude API → Telegram

Make.com scenario for Morning Briefing — daily multi-source aggregation to Telegram

Total execution: ~21 seconds, 10 operations.

The output has six sections: TLDR, WEATHER, YESTERDAY (from session logs), AI SIGNAL (news tied to brand positioning), CONTENT NUDGE (specific content suggestion), and CALENDAR.

Telegram message showing the structured morning briefing output with all six sections

Prompt iterations

First prompt: wall of text, no structure. Round 1 added section headers. Round 2 changed the tone from corporate to conversational. Round 3 gave Claude context about my brands so it connects news to my positioning.

That turned AI SIGNAL from generic news into strategic analysis and enabled the CONTENT NUDGE section.

Failures fixed

Silent failures

Scenario failed with no notification. Added error routes that alert via Telegram.

Partial output

Special characters in Google Sheets broke parsing. Added a text cleaning step.

Error handling complexity

Make.com retry logic and fallback routes took multiple iterations to get right.

Result

Runs every day at 08:00, zero manual input. Under 2 minutes to read. Replaced 10–20 minutes of tab-hopping. ~133 Make.com credits per week. Running in production daily without interruption.